Continuous stress monitoring offers a potential to help understand different mental stress patterns and how clinical intervention could best be applied. One economical way to detect stress is to measure galvanic skin response (GSR) as the electrical conductance of skin varies with physiological arousal. In this work, we studied the effects of different activities (sit, stand and walk) on GSR measurements. We implemented a GSR sensor system and an activity recognition system. We showed that using two accelerometers (at thigh and ankle each) achieved an overall accuracy of 94.7% in activity recognition, an improvement of +27.3% from using single sensor node system. We further demonstrated that the activity information could help improve the sensitivity in stress detection at sitting and standing positions.